Why enterprises should look to open source to solve their observability challenges

Marck Robinson Profile picture for user Marck Robinson October 25, 2021
With rising adoption of microservices and 5G on the way, technology observability has a new level of complexity. Marck Robinson of New Relic explains why it needs to be open, and how we get there.

Today’s enterprises face an overwhelming degree of technological complexity — a web of data, applications and software services that make it difficult to understand what’s happening in your tech stack on a good day, and nearly impossible to fix quickly when things go wrong.

Two major technology trends promise to make things more complicated in the near future. The first is the growing adoption of microservices, single-function, often proprietary tools that address a specific need within the software stack; the second is the coming arrival of 5G, which promises to dramatically increase the speed and scale at which organizations must process data.

In the face of all this complexity, many companies have incredible difficulty with observability — understanding the state of their tech systems at any moment in time. Most engineers are not observability experts, and even when an enterprise uses a well known observability service for their tech stack, they often don’t know how to get started and make the most of their tools. Figuring out how to monitor, debug, identify problems and solve them quickly is difficult and time-consuming when starting from scratch.

What’s the solution? As a community, we need to make observability open. By using open agents, open source tools and an open community of best practices and tools, we can create an open hub where the world’s leading experts and technologies are brought together to help engineers get started and finally understand what’s happening in their systems.

Why open observability?

Open observability already has a tremendous starting place with OpenTelemetry. These open source instrumentation tools make it easier for engineers to collect and measure their telemetry data, an important first step for understanding system performance. However, open observability provides the tools and expertise needed for organizations to collect, organize and understand data from all sources, not just telemetry data. By working towards a unified platform with the combined expertise of our industry, we can help organizations gain a complete, end-to-end picture of their tech stack.

According to New Relic’s 2021 Observability Forecast, a majority of IT decision-makers are familiar with observability; however, there’s a gap in their observability practices. Only 26% of respondents report having a mature observability practice in place, with the most commonly cited barriers to success being lack of resources (38%) and skill gaps (29%). We can make this easier on engineers if we work together.

How we get there

No two enterprise tech stacks are alike. True open observability will require an all-hands-on-deck approach from industry leaders to provide a comprehensive set of tools for organizations to unlock value from their data and make it part of their daily, data-driven approach to engineering across the product lifecycle. If we can provide a large enough collection of building blocks that covers widely-used clusters and frameworks like Gatsby and Kafka, we can put engineers in a much stronger position to instrument their data and learn how to act on it. For our part at New Relic, we are embracing an open approach; between our support for OpenTelemetry and our work to make Pixie an open standard for Kubernetes observability, we are working to make these tools as accessible and understandable as possible to all.

Engineers are facing a tipping point with regard to complexity. If observability leaders commit to open source, we could see thousands of telemetry sources, alert quickstarts and pre-built dashboards made available by the engineering community. With a rich and deep enough repository, millions of engineers throughout the world could use these tools to make better decisions throughout the lifecycle of their software.

We’ve seen what happens when SaaS providers lean too heavily on proprietary models – we end up with something akin to the current situation with microservices. These tools largely do an exceptional job on an individual level, but they’re not designed to play well together. When faced with a challenge like understanding your entire tech stack, openness and interoperability are fundamental needs.

Today’s enterprises need observability, and today’s observability needs open standards.

A grey colored placeholder image